10281157214-ai-toolkit-alpha

What's Changed

• Remove DataParallel for preview version by @vriveras in #85
• Adding base only inferencing by @vriveras in #87
• Moving Phi-2 to HF provider by @vriveras in #89
• Fixing true to True by @vriveras in #91
• New style of prompt templates by @rodrigovaras in #101
• Rename type to task by @kuojianlu in #102
• feat:Adding Phi3 initial template for local fine-tuning by @vriveras in #103
• feat: add inference => finetuning project template name mapping by @a1exwang in #104
• fix:Fixing prompt template for Phi3 by @vriveras in #105
• fix: Missed one end tag and removing a new line by @vriveras in #106
• feat: Adding Phi3 inference models for 128k context window. by @vriveras in #108
• Adding new DML Index for models while we have the official one. by @vriveras in #110
• fix: remove huggingface secret in bicep by @XiaofuHuang in #111
• feat: increase GPU instance number by @XiaofuHuang in #107
• fix: update aca image by @swatDong in #112
• feat: Add inference model metadata by @a1exwang in #109
• fix: hf download and image by @swatDong in #113
• feat: add remote support to all template by @XiaofuHuang in #115
• fix: update replica timeout to 3 hours by @XiaofuHuang in #116
• fix: move timeout to be a parameter by @XiaofuHuang in #117
• fix: add default location in bicep parameters by @XiaofuHuang in #118
• Update olive configs to match new format by @kaysabelle in #114
• docs: add remote development into readme by @XiaofuHuang in #119
• fix: for remote finetuning, add step to check finetune output by @swatDong in #121
• fix: update resource prefix by @XiaofuHuang in #122
• docs: add remote log to readme by @swatDong in #123
• Add phi-3 fine-tuning mapping by @kuojianlu in #125
• Adding temporary olive wheel until release by @vriveras in #124
• docs: add catalog into README by @XiaofuHuang in #127
• docs: update readme file with the latest UI changes and comments by @XiaofuHuang in #126
• feat: adding cuda variant back for linux build by @vriveras in #128
• docs: add private feature flag in remote development by @XiaofuHuang in #129
• Add templates for Meta-Llama-3-8B by @kaysabelle in #120
• Remove Olive wheels in favor of olive-ai version 0.6.0 by @kaysabelle in #131
• Remove wandb and update transformers across all model templates by @kaysabelle in #132
• fix: update remote config according to the latest deps version by @XiaofuHuang in #133
• fix: specify inference dependency version in command by @XiaofuHuang in #134
• fix: decrease startup probe period from 4 min to 2 min by @XiaofuHuang in #135
• fix:moving to current official list of models by @vriveras in #136
• fix: update phi-2 model version in remote development by @XiaofuHuang in #137
• fix: change adapter path by @XiaofuHuang in #138
• fix: add mistral model playground and finetune mapping by @XiaofuHuang in #139
• fix: adapter path according to model by @XiaofuHuang in #140
• feat:adding mistralai/Mistral-7B-Instruct-v0.2 template by @vriveras in #141
• Update mistral model in finetuning command by @kaysabelle in #142
• feat: add OpenAI compatible inference API by @a1exwang in #146
• feat: add OpenAI compatible inference API for other templates by @a1exwang in #147
• Release/v0.3.4 by @a1exwang in #148

New Contributors

• @kuojianlu made their first contribution in #102
• @a1exwang made their first contribution in #104
• @XiaofuHuang made their first contribution in #111
• @swatDong made their first contribution in #112
• @kaysabelle made their first contribution in #114

Full Changelog: https://github.com/microsoft/windows-ai-studio-templates/commits/10281157214-ai-toolkit-alpha